Skip to content
IEMI Labs

SmokeSentinel AI

A solar 360° wildfire smoke camera you can afford to put everywhere

Target price from$4,900per node incl. solar kit

Software: $199–$349 per month per node

Concept, not for sale yet. Tell us if you want it; that decides what we build.

Rate this idea and tell us what to change
SmokeSentinel AI concept: the idea and where each part goes

Drag to rotate · right-drag to move · scroll to zoom · click a part for its size and placeSwipe to rotate · pinch to zoom · two fingers to move · tap a part for its size

How it fits together

SmokeSentinel AI, part by part

Scroll or tap a part. Press “See the 3D CAD design” to watch the model, built from our CAD, light each part up and show how it connects.

SmokeSentinel AI concept: the idea and where each part goes

Real-time 3D render · drag to rotate · final design may varySwipe sideways to rotate

  1. 01

    The whole SmokeSentinel AI

    Cameras on a mast scan the horizon for smoke. The compute module checks each frame; a cellular modem sends the bearing to the owner.

  2. 02Brain

    Raspberry Pi CM5

    The brain: it runs the smoke orchestrator program

  3. 03

    Four wide cameras

    Take 12 MP stills of the horizon, one per compass side

    Raspberry Pi CM5 · CSI-2 ribbon cable · 12 MP still frames

  4. 04

    Hailo-8L AI module

    Runs the smoke detector on every image tile

    Raspberry Pi CM5 · PCIe (M.2 slot) · Image tiles and detections

  5. 05

    SHT45 humidity sensor

    Reads air temperature and humidity in a radiation shield

    Raspberry Pi CM5 · I²C (3 m shielded lead) · Temperature and humidity

  6. 06

    nRF9151 cellular modem

    Wakes the CM5, sends alerts by LTE-M or satellite

    Raspberry Pi CM5 · UART, 1 Mbaud · Alerts, crops and commands

  7. 07

    5 V step-down regulator

    Powers the CM5 only when the modem switches it on

    Raspberry Pi CM5 · 5 V power lead · 5 V, 5.5 A

  8. 08

    All connected

    6 key parts, one program, and your phone

What it is

A solar-powered four-camera head on a mast that watches the horizon for wildfire smoke.

What it does

  • Scans 360° every minute
  • Detects smoke plumes on-device
  • Alerts over LTE or satellite
  • Locates fires with neighbouring nodes

Who it's for

Rural councils, utilities, co-ops, ranches and wildland–urban communities.

The idea

Solar camera nodes on a ridge and a water tower spot wildfire smoke and send the owners a bearing and a photo

The problem today

On a red-flag afternoon a pasture fire on a big ranch can burn for many minutes before anyone sees the smoke. Agency camera networks cost about $50,000 a station a year and rarely cover private ranch land, so owners rely on someone happening to look.

With SmokeSentinel AI

A SmokeSentinel node on the ridge watches the skyline all day and sends the owner group a bearing and a snapshot over LTE-M, or a text over satellite where there is no signal. A second node on the water tower crosses its bearing so the app shows where the plume is. People then call 911; the node never does.

Fence lineNorth pastureHay pastureWooded creekRidgeRanch houseWater towerSmoke plumeCounty road123
  1. 1SmokeSentinel Ridge node on the ridge mastThe brain: solar, two cameras over about 200 degrees, sees a plume and sends a bearing and snapshot over LTE-M or satellite
  2. 2SmokeSentinel 360 node on the water towerFour cameras all round; when it sees the same plume its bearing crosses the ridge node's to place the fire on the map
  3. 3Owner group phones at the ranch houseOne app: the alert, the photo and the plume on the map; a person decides and calls 911, the node never dials it
A Texas Hill Country ranch: pastures, a wooded creek, a ridge, the water tower and the ranch house, drawn to show where each part goes (not to scale).

How the parts connect

SmokeSentinel AI, wired up

Every part, what links it to the brain, and how the data reaches your phone. A solid line is a cable; a dashed line is a radio link.

Four wide cameras12 MP still framesCSI-2 ribbon cableHailo-8L AI moduleImage tiles and detectionsPCIe (M.2 slot)SHT45 humidity sensorTemperature and humidityI²C (3 m shielded lead)nRF9151 cellular modemAlerts, crops and commandsUART, 1 Mbaud5 V step-down regulator5 V, 5.5 A5 V power leadTHE BRAINRaspberry Pi CM5runs smoke orchestratorNode.js with libcamera and Hail…IEMI cloudvia LTE-M · MQTT/TLSOwner's phonevia Push, SMS or voice call
Drawn from the SmokeSentinel AI parts list. The 3D view above is built separately, from the CAD.

The program

What controls SmokeSentinel AI

Runs on
Raspberry Pi CM5
Written in
Node.js with libcamera and HailoRT
System
Raspberry Pi OS Lite (64-bit)
AI model
YOLO smoke detector on Hailo-8L
  1. 1Take a 12 MP still from each camera
  2. 2Crop the horizon band and cut it into 640 px tiles
  3. 3Run the smoke detector on the Hailo-8L
  4. 4Track candidates: three frames, two minutes, growing 20%
  5. 5Send the alert and a small crop to the modem
smoke orchestrator.tsmain loop
for (;;) {
  for (const cam of cameras) {
    const still = await libcameraStill(cam.id)           // AF locked at infinity
    const band = cropBand(still, cam.horizonMask)        // about 4608 x 900
    const hits = await hailo.infer(tile(band, 640))      // smoke class
    tracker.update(cam.id, hits)                         // IoU + growth + persistence
  }
  for (const c of tracker.confirmed()) {
    if (inMuteZone(c) || sunGlare(c)) continue
    modem.write(alertJson(c, cropJpeg(c)))               // UART to nRF9151
  }
  await sleep(fireWeatherMode ? 30_000 : 60_000)
}

To your phone

How the data reaches SmokeSentinel map and alerts

Phone push, SMS, voice call and web map

  1. Mast head
  2. IEMI cloud
  3. Owner's phone

What you see in the app

One screen for each job SmokeSentinel AI does.

  • Smoke spottedA thin plume rises behind the west ridge on a dry afternoon. The node checks it keeps growing, then sends a snapshot to the ranch.

  • Solar and batteryIn midsummer the node sits on its mast in full sun with no mains power. The hourly report shows its battery and head heat.

  • Node networkA co-op runs twelve nodes across the valley. Each reports every hour, so one that goes silent stands out.

App screens with sample data. The interface is in development.

What the app gets

Data SmokeSentinel AI sends to the app
Smoke alertBearing 268°, 5-7 kmEach alert
Alert crop640×360 JPEGEach alert
Battery13.2 V, 88%Hourly
Node healthHead 41.5 °C, RH 19%Hourly
SignalRSRP −108 dBmHourly

Outbound-only link with no open ports. Night images are kept on the node unless the owner sends them.

CAD design · Design complete (stages 1–6), not built

Designed in code, checked in 3D

  • 9 printed parts, each a single closed solid
  • Every bought part fitted inside and checked for collisions
  • Assembled exactly as shown in the 3D CAD view (the button on the picture at the top)
See the CAD parts renderSmokeSentinel AI printable parts rendered from the CAD

Key specs

SmokeSentinel AI hardware
AI processorHailo-8L (13 TOPS) on a low-power ARM host, duty-cycled
Camera4× 5 MP low-light cameras (Sony STARVIS 2 class), ~100° each for 360° coverage
SensorsTemperature and humidity sensor · GNSS for location and time · Tilt / tamper sensor
Connectivity4G LTE (Cat 1 / LTE-M) · Satellite fallback for alert messages (Myriota HyperPulse or Semtech LR2021 S-band class) · Optional LoRaWAN or Wi-Fi HaLow for local mesh
Power25 W solar panel + 12 V 18 Ah (230 Wh) LiFePO4 with MPPT (Voltaic CORE-class kit); ~3 W average duty-cycled (target)
EnclosureMast-mounted cylindrical camera head with four camera windows and sun hood; separate battery box; 25 W solar panel
DimensionsHead 30 cm Ø × 18 cm; 25 W solar panel; battery box
WeightHead ~3.5 kg; battery box ~4 kg
Ingress ratingIP66
All target specifications

Detection

Detection target specifications
Field of view360° horizontal
Detection rangeUp to ~10 km in clear conditions (target)
Scan interval30–60 s, adaptive
Time to alertUnder 3 min from visible smoke (target)

Connectivity

Connectivity target specifications
Primary4G LTE Cat 1 / LTE-M
FallbackSatellite short messages
Image uploadCompressed snapshots over LTE
APIsREST, webhooks, CAP alerts (planned)

Power and physical

Power and physical target specifications
Solar25 W panel
Battery230 Wh LiFePO4, ~3 days autonomy
Head size30 cm Ø × 18 cm
Operating temp.−20 °C to 60 °C

Targets for production units; they may change during development.

Pricing

Hardware: $4,900–$6,900 per node incl. solar kit

Software: $199–$349 per month per node

Target pricing. Subscription includes LTE data, portal and model updates; satellite messaging is billed at cost. Mast and installation quoted per site.

Indicative launch pricing in USD, excluding taxes and shipping. Final pricing is confirmed before any balance is charged.

Community

What do you think of SmokeSentinel AI?

SmokeSentinel AI is in development. Rate the idea, tell us what to change about the features, specs or price, and we'll build it with your feedback.

How much people want it

–

No ratings yet

  1. 50
  2. 40
  3. 30
  4. 20
  5. 10

Follow SmokeSentinel AI

Get build updates and first access when pre-orders open.

Leave feedback

Your rating of the idea

Feedback on a product concept in development, not a review of a shipped product. Posts may be moderated; links and contact details are held for review. See our community guidelines and privacy policy.

Comments

    No comments yet. Tell us what you'd want from SmokeSentinel AI.

    Questions

    Ask about SmokeSentinel AI

    Tell us what this would save you, or ask anything. We reply within two business days.

    We use your details only to respond to this request. See our privacy policy.

    From the product lab · ideation phase

    Lab updates for SmokeSentinel AI

    All lab updates
    1. LatestRanches, property-owner groups and rural co-ops in fire country often have no camera watching their land, while agency smoke-camera stations cost about $50,000 a station a year (SmokeSentinel DESIGN.md market check, TechCrunch and MIT Technology Review).
      • Drawn idea plan (idea.json): a Hill Country ranch from above, with its pastures, wooded creek, ridge, water tower, ranch house and county road; pins for the Ridge node on its mast, the 360 node on the water tower and the owner group phones at the house, with the two bearings crossing at a plume
      • The product picture is now this drawn plan instead of the plain fallback card; drawn from the design notes, not the CAD
      • What the plan keeps from the market work: the nodes are a gap filler that tips off people; a person calls 911 and the node never dials it, and it never replaces agency cameras or evacuation orders
      • Next: close-out C9 (recommend the first product to the owner) from 30 Oct 00:00; until then more drawn idea plans for catalogue products

      See it in 3DFull research and design

    2. Fruit and wine growers next to woods face three jobs in the same summer: smoke from a fire in the hills, pests to count every week, and crews working in heat under California rules (shade from 80 °F, Cal/OSHA §3395; act on smoke from AQI 151, §5141.1). Agency smoke cameras cost about $50,000 a year, camera traps are sold per trap per season, and a Kestrel heat meter is about $554; our own three products for it cost $5,898 and $224 a month.
      • Combined into FieldEdge (https://iemilabs.com/products/fieldedge/); SmokeSentinel AI itself is unchanged

      See it in 3DFull research and design

    SmokeSentinel AI FAQ

    General FAQ
    Does it replace long-range tower cameras?

    No. SmokeSentinel is designed for short-to-mid-range infill and for places that cannot justify a tower system. It complements existing networks and fire-agency detection, and its alerts can feed the same portals.

    Is image data processed on the device?

    Yes. Smoke detection runs on the node. Only alerts with compressed snapshots, and periodic low-resolution health images, are sent. Cameras look at the horizon, not at people, and no identity data is processed.

    How is it installed?

    The head, panel and battery box mount on a 4–10 m mast, pole or rooftop. No power or network cabling is needed. A two-person crew can install a node in about half a day.

    What connectivity does it need?

    SmokeSentinel uses 4G LTE where available. Where there is no coverage, short alert messages are sent by satellite; images are uploaded when LTE returns or through a local mesh link.

    What happens to our data and models if we leave?

    You can export detections, images and the site-tuned model weights at the end of your contract. We do not lock customers in.

    Is SmokeSentinel AI available now?

    No. It is a new concept in development. Specifications are target specifications. A 2-year limited hardware warranty is planned. It is a concept: we are not taking orders, reservations or deposits, and it has no ship date yet.

    Related products

    RaidPost concept: the idea and where each part goes
    VisionConcept

    RaidPost

    A solar pole head with a camera, strobe and siren, plus a bell for the house.

    • Names the animal in the gap
    • Fires a random light and siren
    • Rings a bell in the house

    From $145 per post

    Details
    FlowCount AI concept: the idea and where each part goes
    VisionConcept

    FlowCount AI

    A small white ceiling sensor with two lenses that counts people without sending images.

    • Measures queue length in real time
    • Predicts wait times per lane
    • Counts entries, exits and occupancy

    From $899 per sensor

    Details
    AirfieldWatch AI concept: the idea and where each part goes
    VisionConcept

    AirfieldWatch AI

    A thermal and visible camera turret on a mast that spots birds and wildlife near runways.

    • Detects birds and flocks day and night
    • Classifies hazard-relevant species groups
    • Triggers dispersal devices automatically

    From $8,900 per turret

    Details

    SmokeSentinel AI

    Concept · target $4,900

    I want this